GoldMiner

Viewing 15 posts - 1 through 15 (of 16 total)

1 2 →

Author

Posts
02/24/2020 at 5:36 PM #120331 quote

Francesco
Participant

Veteran

Noticed that there are no systems on Gold, so i tryed to develop one by myself.
Here’s what i got, simple and effective

Based on a simple moving average and a Breakeven function, with a limit to the operative time.

Works on 4H.

Paul, nonetheless and Nicolas thanked this post

GoldMiner1.jpg GoldMiner2.jpg GOLD4HL.itf

02/24/2020 at 5:48 PM #120336 quote

robertogozzi
Moderator

Master

You should also post the whole code so that some interested may take a look at it without having to import it.

Thank you. 🙂

02/24/2020 at 5:53 PM #120337 quote
Francesco
Participant

Veteran
Usually didn’t seen anyone doing that, but here it is 😉
```
defparam cumulateorders = false

// Long
If average[20] CROSSES OVER average[55] THEN
BUY 0.2 CONTRACTS AT MARKET
ENDIF

// Stop and target
SET STOP pLOSS 0
SET TARGET pPROFIT 25

IF OnMarket AND (BarIndex - TradeIndex) >=93 THEN
SELL      AT MARKET
EXITSHORT AT MARKET
ENDIF

StartBreakeven = 0 // How much pips/points in gain to activate the Breakeven function?
PointsToKeep = 20 // How much pips/points to keep in profit above of below our entry price when the Breakeven is activated (beware of spread)

// Reset the BreakevenLevel when no trade are on market
IF NOT ONMARKET THEN
BreakevenLevel=0
ENDIF

// Test if the price have moved favourably of "startBreakeven" points already
IF LONGONMARKET AND close-tradeprice(1)>=startBreakeven*pipsize THEN
//Calculate the BreakevenLevel
BreakevenLevel = tradeprice(1)+PointsToKeep*pipsize
ENDIF

// Place the new stop orders on market at BreakevenLevel
IF BreakevenLevel>0 THEN
SELL AT BreakevenLevel STOP
ENDIF

IF SHORTONMARKET AND tradeprice(1)-close>startBreakeven*pipsize THEN
//Calculate the BreakevenLevel
BreakevenLevel = tradeprice(1)-PointsToKeep*pipsize
ENDIF

//Place the new stop orders on market at BreakevenLevel
IF BreakevenLevel>0 THEN
EXITSHORT AT BreakevenLevel STOP
ENDIF
```
robertogozzi thanked this post
02/24/2020 at 6:21 PM #120339 quote

robertogozzi
Moderator

Master

You may have noticed code is usually not posted on large topics because people already know know what’s it all about and posting code everytime would make the topic difficult to read, but the first time it’s better to let the reader decide whether it is worthwhile or not (for him) importing it.

Thank you.

Francesco and jebus89 thanked this post

02/24/2020 at 7:10 PM #120342 quote

Francesco
Participant

Veteran

I was talking about first posts of new topics that i’ve seen recently, but doesn’t matter, it was what it was 🙂
It’s surely more smart put the entire code in the first post for the reasons you listed.

02/24/2020 at 7:18 PM #120343 quote

Vonasi
Moderator

Master

I always like to see the code in the first post as I don’t always have my platform open 24/7 and so seeing how the strategy is coded is a lot of hassle of opening the platform and importing the code and I am generally to busy to bother doing that. If the code is in the first post however then I am able to immediately check it out and see if it is of interest or if I have anything to comment that might assist.

02/25/2020 at 12:28 AM #120360 quote

nonetheless
Participant

Master

Small alteration to your exit conditions:

IF OnMarket AND (BarIndex – TradeIndex) >=93 and close>positionprice THEN …

Makes a huge difference on 100k — gain/loss ratio of 370? worst trade -$4? seriously?

Probably worth re-optimizing the bar number.

02/25/2020 at 8:19 AM #120372 quote

Vonasi
Moderator

Master

nonetheless wrote:

Makes a huge difference on 100k — gain/loss ratio of 370? worst trade -$4? seriously?

Yes but one day you will open a trade at the top of the market and then the market will take a dive and you will go into massive account killing drawdown! Just because it does not happen in your backtest does not mean it can’t happen. Perhaps a robustness test would show this as perhaps the big DD trades are just not happening because of the start date and time meaning that you just so happened to be on the market and so missed those bad trades.

02/25/2020 at 9:38 AM #120386 quote

nonetheless
Participant

Master

Yes, the tone of voice was meant to imply that this may well be too good to be troo and needs more testing … it’s effectively only closing if in profit. But may be worth considering …

02/25/2020 at 10:18 AM #120394 quote
Paul
Participant

Master
added re-entry to see what it does, kind of a robustness-test
```
defparam cumulateorders = false

once reentry              =1 // close & re-entry on signal in same direction

condbuy=average[20] CROSSES OVER average[55]

if condbuy then
signal=1
else
signal=0
endif

graph signal

ctime=1

// Long
if ctime then

If condbuy THEN

if not onmarket then
buy 1 contract at market
endif

if reentry then
if longonmarket then
sell at market
endif
endif
endif

if reentry then
if condbuy[1] then
if longonmarket[1] then
buy 1 contract at market
endif
endif
endif

endif


// Stop and target
SET STOP pLOSS 0
SET TARGET pPROFIT 25

IF OnMarket AND close>positionprice and (BarIndex - TradeIndex) >=1 THEN
SELL      AT MARKET
ENDIF

StartBreakeven = 0 // How much pips/points in gain to activate the Breakeven function?
PointsToKeep = 20 // How much pips/points to keep in profit above of below our entry price when the Breakeven is activated (beware of spread)

// Reset the BreakevenLevel when no trade are on market
IF NOT ONMARKET THEN
BreakevenLevel=0
ENDIF

// Test if the price have moved favourably of "startBreakeven" points already
IF LONGONMARKET AND close-tradeprice(1)>=startBreakeven*pipsize THEN
//Calculate the BreakevenLevel
BreakevenLevel = tradeprice(1)+PointsToKeep*pipsize
ENDIF

// Place the new stop orders on market at BreakevenLevel
IF BreakevenLevel>0 THEN
SELL AT BreakevenLevel STOP
ENDIF

IF SHORTONMARKET AND tradeprice(1)-close>startBreakeven*pipsize THEN
//Calculate the BreakevenLevel
BreakevenLevel = tradeprice(1)-PointsToKeep*pipsize
ENDIF

//Place the new stop orders on market at BreakevenLevel
IF BreakevenLevel>0 THEN
EXITSHORT AT BreakevenLevel STOP
ENDIF
```
the breakeven though should be the other way around with it’s value’s

when reaching a gain, then move breakevenlevel to 0, however it’s a nice equitycurve this way.

Screenshot-2020-02-25-at-10.12.24.jpg Screenshot-2020-02-25-at-10.12.51.jpg
02/25/2020 at 10:56 AM #120400 quote

Vonasi
Moderator

Master

nonetheless wrote:

Probably worth re-optimizing the bar number.

This is a terrible idea. Optimising on any time based idea is always going to be a great way to curve fit. As soon as I see numbers like 93 it instantly gets me worried that it is totally curve fitted.

02/25/2020 at 11:51 AM #120415 quote

nonetheless
Participant

Master

Vonasi wrote:

As soon as I see numbers like 93 it instantly gets me worried that it is totally curve fitted.

But surely there must always be some optimal number, no? where optimal does not necessarily mean highest profit, could be lowest drawdown or %win or whatever helps you sleep at night.

I don’t think there’s a good ‘theoretical’ reason for it not to be 93 or 71 or 57, esp in a 10 year backtest with WF and all the rest. Or is there?

02/25/2020 at 12:17 PM #120419 quote
fifi743
Participant

Master
```
defparam cumulateorders = false

once reentry              =0// close & re-entry on signal in same direction

condbuy=average[20] CROSSES OVER average[55] and (average[55]-average[55](close[1]))/pointsize> 0.1

if condbuy then
signal=1
else
signal=0
endif

graph signal

ctime=1

// Long
if ctime then

If condbuy THEN

if not onmarket then
buy 1 contract at market
endif

if reentry then
if longonmarket then
sell at market
endif
endif
endif

if reentry then
if condbuy[1] then
if longonmarket[1] then
buy 1 contract at market
endif
endif
endif

endif


// Stop and target
SET STOP pLOSS 0
SET TARGET pPROFIT 25

IF OnMarket AND close>positionprice and (BarIndex - TradeIndex) >=1 THEN
SELL      AT MARKET
ENDIF

StartBreakeven =0// How much pips/points in gain to activate the Breakeven function?
PointsToKeep = 20 // How much pips/points to keep in profit above of below our entry price when the Breakeven is activated (beware of spread)

// Reset the BreakevenLevel when no trade are on market
IF NOT ONMARKET THEN
BreakevenLevel=0
ENDIF

// Test if the price have moved favourably of "startBreakeven" points already
IF LONGONMARKET AND close-tradeprice(1)>=startBreakeven*pipsize THEN
//Calculate the BreakevenLevel
BreakevenLevel = tradeprice(1)+PointsToKeep*pipsize
ENDIF

// Place the new stop orders on market at BreakevenLevel
IF BreakevenLevel>0 THEN
SELL AT BreakevenLevel STOP
ENDIF

IF SHORTONMARKET AND tradeprice(1)-close>startBreakeven*pipsize THEN
//Calculate the BreakevenLevel
BreakevenLevel = tradeprice(1)-PointsToKeep*pipsize
ENDIF

//Place the new stop orders on market at BreakevenLevel
IF BreakevenLevel>0 THEN
EXITSHORT AT BreakevenLevel STOP
ENDIF
```
Hello,
Or add a filter

Paul thanked this post

GOLD4HL-1.itf Capture-d’écran-639.png
02/25/2020 at 2:00 PM #120443 quote

Francesco
Participant

Veteran

Robustness test results, nothing to do here 🙁

Thanks to @paul for the excel file

1-4.jpg 2-3.jpg 3-2.jpg

02/25/2020 at 3:52 PM #120469 quote

Vonasi
Moderator

Master

nonetheless wrote:

But surely there must always be some optimal number, no?

Yes – but that optimal number is your curve fit. You have looked at history and said ‘that number of bars before exiting has worked the best in the past’ but that does not mean that it will work tomorrow or ever be the best number to use ever again. Now if you had say a range of numbers from 83 to 103 that all worked pretty similarly and all returned nice equity curves then picking 93 slap bang in the middle of that range might put you in with a fighting chance that it will continue to work even if 93 was not the best result or best equity curve.
Author

Posts